I investigated the longitudinal relationship between alcohol use, cannabis use and cigarette smoking and serious violence using data from a prospective longitudinal, nationally representative cohort of 17,519 individuals interviewed on 4 occasions. Participants were between 12 and 17 years of age at wave I and were between 23 and 32 year of age at wave IV. There was a linear relationship between the number of drinks consumed but not frequency of drinking alcohol and violence. The number of individuals needed to abstain from drinking alcohol to prevent one from becoming violent was estimated. Smokers were also twice as likely to report subsequent violence within the next year, however there was no relationship between cannabis use and incident violence within the next year. Analysis of the entire cohort (whether or not they reported violence at baseline) incorporated individual change in substance use over time to investigate the longitudinal relationship between substance use and violence. Moderate drinkers were approximately 1.4 times more likely to be violent than non-drinkers. Cigarette smoking and cannabis use was also associated with similar increases, but heavier drinkers were more than twice as likely to report serious violence. When the trajectories of violence were investigated, predictive marginal effects showed that drinking 1-4 drinks on each occasion was associated with an increased risk of violence during adolescence, but as the individuals got older the risk of violence tended to converge with that of non-drinkers by age 20. Heavier drinking however, was associated with a greater risk of violence well into adulthood, although the trajectories tended to converge by age 30. The relationship between personality traits, alcohol and violence was then investigated using structural equation modelling. Alcohol was found to mediate the association between violence and specific personality characteristics, especially anger/hostility and extroversion.